Can we replace the Tiwai smelter with a data center?

(I was going to save this for Nine To Noon tomorrow, where I’m doing the technology segment, but things are moving fast. But do listen in)

A few people have muttered, sometimes out loud, that one possibility for the apparently uneconomic Tiwai point aluminium smelter is to switch it to become a data center.

So does it make sense?

Well – aluminium smelters use a lot of electricity. The Tiwai Point owners have a contract for electricity supply with Meridian Energy for the continuous supply of 572 megawatts for the period 2013 to 2030. The first three years from Jan 1, 2013 are locked in – a so-called take or pay arrangement. (Apparently, according to Chalkie, NZAS buys on the spot market to make a total of 625MW of demand.)

But let’s stick with 572 megawatts. That’s a lot, and multiplying it out by 24 hours a day and 365 days a year shows that the smelter uses about 5,000 gigawatt hours per year from the Meridian contract.

It’s no mistake that the annual electricity output from the Manapouri dam is about the same as the demand from Tiwai – 5,100 GigaWatt hours. It’s a sizeable power plant, not the biggest in New Zealand (Huntly is bigger), but certainly up there with some of the world’s largest.

The average NZ Household uses around 8,000 kilowatt hours per year. So the power supplied to Tiwai could supply about 626,340 houses each year – out of a total of around 1.6 million residential connections in NZ. This is a big piece of the electricity supply equation.

So how much does a data center use? Well the new IBM one in Auckland, which cost $80m, has backup generator capacity for “266 homes”. So to absorb Tiwai’s electricity use we’d need we’d need 2,354 of those flash Auckland IBM data centers, at a total cost of $188 billion.

That’s not going to happen.

Meanwhile Tiwai is at the end of the earth from America – in a country with only one cable system connecting it to the world. There is no compelling reason to house a decent sized data center in Invercargill.

Data centers do like power (and the cold), but not nearly as much as aluminium smelters.

The NYTimes reported, in September 2011, that Google itself uses a continuous 260 MegaWatts of electricity worldwide. They compared this to a 60 watt bulb being used for 3 hours per month per user.

Impressive as it sounds, all of Google’s demand for data center power is still less than half the freed up capacity from Tiwai.

So no – we cannot replace Tiwai’s electricity demand with a data center. Or even 2,354 of them.

Meanwhile Tiwai is a brownfields site – we’d need to remove the contaminated carbon pots used to make the aluminium, decontaminate and decommission the buildings and plant, including the carbon plant where the anodes are made and baked, and so on. It’s going to cost hundreds of millions of dollars. The pots themselves have a limited life, and one smart thing to do could be to scale down production for as you take pots offline. However at some point it will be to hard to run and uneconomic, and it would be best to simply tap the aluminium from each pot in turn and switch the plant off. That’s a big job, by the way. 

Some aluminium economics

Tiwai smelter is the 28th largest in the world, and slipping down the list as larger plants are built. I’ve worked or been in three of the others on that list, two larger that Tiwai and one smaller. The smaller one closed half of their plant due to ongoing electricity supply problems, and because it was less economic than the more modern larger ones. And that’s how it works – smaller and older plants are inherently less cost effective than the newer larger ones, and while differences in operating effectiveness can help, ultimately the underlying technology, such as the size of the pots, dictates the effectiveness at converting electricity and alumina into aluminium. Smart operators set the price of their alumina and electricity inputs as a function of the aluminium price, and so the plant efficieny dominates the equation.

Tiwai is now a third of the size of the world’s largest plants, and will certainly be higher cost than many on the list – the newer ones, but not the highest.

RUSAL, which owns the largest two plants which produce over 1 million tonnes a year each, and has 9% of the global market, reckons that 1 to 1.5 million tonnes worth of annual production from the industry would be idled this year. That means supply is exceeding demand, and the high cost producers are the ones that get hit first.

However other countries and operators may not understand that their plants are the highest cost, and they may be subsidising the plant. This is not a fight that tiny New Zealand stands to win, if we enter it, ans so the right thing to do is hold firm on the pricing of electricity, and let the market decide.

But it all comes back to the rising demand from China – perhaps a boom year there could see demand fro aluminium spring up to absorb the industry capacity. Or not. It’s not a game for the faint hearted, but the owners of the higher cost smelters know that shutting them down will reduce global supply, and doing that will increase the price for all their existing smelters.

Published by Lance Wiggs